Director of the JBA Trust and Lancaster University Professor in Practice, Rob Lamb, a hydrologist and numerical modeller, gave an enlightening presentation at Lancaster University regarding the uncertainties involved with the modelling of floods as well as other environmental risks. After an appreciation of the risk flooding posed to the United Kingdom and its increasing cost and relevance, Rob went into the science behind measuring, predicting and modelling such events.
Points raised included the lack of ‘on the ground’ data for Natural Flood Management, as well as the need for such data on different spatial scales. Without more on the ground data, our models will struggle to accurately represent complex systems such as the environment, therefore accurate data collection in the near future is key to help us understand and combat flood risk.
The ability to accurately measure the impact of certain land use features such as dams, weirs, flumes and infrastructure on flooding was a prominent theme, with each unknown continually adding uncertainty into prediction estimates and therefore reducing their accuracy. The limits and dangers of extrapolation were another strong element of the talk, as well as the considerable impacts uncertainties have upon its reliability. The spatial variability in hydrological properties means we need lots of data for each individual area, as extrapolating between areas or upscaling can quickly start giving us inaccurate projections for flood risk.
The question of stationarity in time-series data (i.e. whether the mean, variance etc. changes with time) was mentioned, with an example of climate change and its impacts on flooding, and the added complexity non-stationarity brings. The need for an increased number of large datasets for many environmental systems was outlined, as well as multiple simulations in an attempt to constrain certain intrinsic uncertainties within the system. Lastly, Rob noted that the output of these models must have a basis in physical science in order to be interpreted and explained, and their uncertainties acknowledged and quantified.
Rob's talk was highly relevant to my research, outlining the need for more data on Natural Flood Management practices which underlie our models. I am extremely grateful to Rob for taking the time to deliver such an interesting talk.
About the Author: Ethan WallaceI graduated from Lancaster University in July 2016 with a First Class Honours in BSc Physical Geography and continued working here as a research assistant developing data science tools for EPSRC until March 2017. My research interests lie in hydrometeorology, hydrology and environmental management, primarily investigating flood prediction and mitigation.